In this talk\, which is based on joint work with Michiel van Lambalgen\, we will extend the analysis of the role of idealisation in linguistic theory that was presented in our 2011 Theoretical Linguistics paper\, to a further investigation of the role that naturalistic assumptions play in shaping linguistic theory. That task has both an ontological and a methodological component. As far as the ontology is concerned we will investigate what role the competence - performance distinction plays in the conception of language as a natural kind\, and what consequences obtain if that distinction is left behind. On the methodological side\, the first task is to get a handle on which questions in linguistics are posed by its object of investigation itself and which are 'merely' the questions that linguistic theories produce.
END:VEVENT BEGIN:VEVENT DTSTAMP;VALUE=DATE:20120517T021507Z DTSTART;VALUE=DATE-TIME:20120530T160000Z DTEND;VALUE=DATE-TIME:20120530T180000Z UID:http://cogsci.uni-osnabrueck.de/en/node/737 URL;VALUE=URI:http://cogsci.uni-osnabrueck.de/en/node/737 SUMMARY:Framing\, scales\, and human rationality DESCRIPTION:There is a widespread rumour in the behavioural sciences that the experimental record shows that people are irredeemably irrational. In this talk I discuss one of the more popular pieces of evidence that have been adduced in favour of this claim\, and argue that it shows exactly the opposite.
Starting with Tversky and Kahneman's Asian Disease study\, psychologists have collected a wealth of experimental data showing that the way a problem is framed may have an effect on people's choices and decisions. Based on a semantic analysis of evaluative expressions like \\"good\\"\, I propose a principled explanation of such framing effects. The key idea is that our choices and decisions reveal a counterfactual systematicity\: they carry information about the choices and decisions we would have made if the facts had been otherwise. It is these counterfactual alternatives that may diverge between otherwise equivalent versions of the same task\, and thus explain the effects of framing.
END:VEVENT BEGIN:VEVENT DTSTAMP;VALUE=DATE:20120517T021507Z DTSTART;VALUE=DATE-TIME:20120606T160000Z DTEND;VALUE=DATE-TIME:20120606T180000Z UID:http://cogsci.uni-osnabrueck.de/en/node/738 URL;VALUE=URI:http://cogsci.uni-osnabrueck.de/en/node/738 SUMMARY:Sequence learning in spiking neural networks DESCRIPTION:Learning to speak\, to sing or to perform specific movements requires the ability to learn stereotypical sequences of movements. Since those sequences of movements are caused by their corresponding sequences of neural activity\, it is of great importance to understand how spatio-temporal spiking patterns can be stored and retrieved in neural networks. Here we consider a biologically realistic recurrent network of stochastic spiking neurons composed of both a hidden and a visible layer. This recurrent network is trained such that after learning the distribution of spatio-temporal spiking patterns produced by the visible network is as close as possible to the target distribution. Importantly\, the learning rule derived from this computational principle is valid for all synapses\, including those that connect visible units to hidden units - which is in contrast with the liquid state machine approach.
END:VEVENT BEGIN:VEVENT DTSTAMP;VALUE=DATE:20120517T021507Z DTSTART;VALUE=DATE-TIME:20120613T160000Z DTEND;VALUE=DATE-TIME:20120613T180000Z UID:http://cogsci.uni-osnabrueck.de/en/node/739 URL;VALUE=URI:http://cogsci.uni-osnabrueck.de/en/node/739 SUMMARY:Indefinites\, questions and methods of identification END:VEVENT END:VCALENDAR